“Downhill” Esophageal Varices in Congenital Heart Disease

Typical esophageal varices occur in the distal esophagus, secondary to portal hypertension, and have been termed “uphill varices” because of the direction of blood flow (1). In contrast, varices of the upper esophagus are rare and have distinct pathophysiology. We report a case of gastrointestinal bleeding secondary to esophageal varices in the upper esophagus in a 14-year-old patient with palliated complex heart disease. This report aims to increase the awareness of this unusual form of esophageal varices secondary to obstruction of systemic venous drainage and treatment implications. A 14-year-old patient with history of complex cyanotic congenital heart disease was palliated with an extracardiac total cavopulmonary anastomosis (Fontan procedure). She had a history of bilateral superior vena cavas (SVCs). Both the right and left SVCs were connected to the respective pulmonary arteries as a Glenn anastomosis. She was admitted to the hospital for investigation of severe recurrent anemia (hemoglobin 4.8 g/dL) for a 2-year period. There was no history of an obvious source of bleeding. The workup ultimately included upper gastrointestinal endoscopy, which revealed dilated tortuous varices in the upper esophagus extending from 15 to 21 cm, measured from the incisors (Fig. 1A). Remarkably, the distal esophagus and the gastric antrum were completely free from varices. This unusual finding prompted further investigations.FIGURE 1: Upper gastrointestinal endoscopy showing (A) dilated tortuous varices in the upper esophagus and near-complete resolution (B).An echocardiogram demonstrated adequate cardiac function without significant valvular regurgitation or stenoses. The Fontan circuit and pulmonary arteries could not be adequately visualized. She was therefore referred for diagnostic cardiac catheterization. This revealed elevated pressure in the Fontan circuit (mean pressure 20 mmHg). The Fontan circuit, the right-sided superior cavopulmonary anastomosis, and the right pulmonary artery were widely patent. A previously placed stent in the left pulmonary artery was patent without stenosis but jailed the entry of the left-sided cavopulmonary anastomosis. There was complete occlusion of the anastomosis of the left vena cava to the left pulmonary artery (Fig. 2A). There was no bridging vein connecting the 2 SVCs, and hence the obstructed left SVC drained via tortuous esophageal venous collaterals, flowing from the proximal to distal esophagus below the diaphragm (Fig. 2B, C).FIGURE 2: Angiogram showing (A) complete occlusion of left superior vena cava (LSVC) to left pulmonary artery connection (arrow), (B) decompressing venous collaterals to the esophageal plexus (arrow), (C) downhill variceal drainage (arrow), (D) widely patent LSVC to left pulmonary artery connection after reestablishing continuity (arrow), and (E) decompressing venous collaterals now predominantly to the left upper pulmonary veins (arrow) without major drainage to the esophageal plexus.Given the recurrent severe blood loss, the cardiology service decompressed the occluded left SVC. A stiff guidewire was passed through the occluded vessel from the left internal jugular vein, and balloon angioplasty was performed. Although a lumen was seen afterward, it was quite irregular and the distal end was jailed by the stent in the left pulmonary artery. A bare metal stent was therefore placed across this area, resulting in an excellent angiographic result (Fig. 2D). She started taking heparin infusion and bridged to anticoagulation with Coumadin. The child was brought back for an elective catheterization 3 months later. During this time, she had no episodes of blood loss and maintained a normal hemoglobin. In spite of therapeutic anticoagulation, angiography revealed that she had reoccluded her left cavopulmonary anastomosis. Fortuitously, she had developed new decompressing collaterals into the left upper pulmonary veins such that the flow of all of the obstructed left SVCs was to the left upper pulmonary vein, bypassing the esophageal plexus (Fig. 2E). This was corroborated by endoscopy, which showed near-complete resolution of the varices (Fig. 1B). The cardiology service was able to recanalize the occluded vessel again, and this time placed a covered stent to minimize in-growth of tissue. She started taking tissue plasminogen activator and reimaged 72 hours later, demonstrating a patent stent. Unfortunately, in spite of therapeutic anticoagulation, the child again developed complete occlusion of the vessel, yet there were no varices on endoscopy, and the occluded vessel continued to drain via the pulmonary veins. In view of this, and in the absence of clinical bleeding, it was decided not to intervene further. DISCUSSION The esophagus has dual venous drainage, such that veins from the lower esophagus anastomose with the coronary vein, which then flows into the portal system, whereas its upper portion drains via a number of systemic veins eventually into the SVC (1,2). Esophageal varices are typically secondary to portal hypertension and occur at the lower end of the esophagus (1). Lesser known, but increasingly recognized, are varices that involve the upper esophagus. These occur secondary to obstruction of the SVC. These upper esophageal varices represent dilated tortuous veins draining blood away from the obstructed SVC to the portal system in a downward direction, and are dubbed “downhill” varices (1,2). Our patient demonstrated downhill varices in the upper esophagus secondary to otherwise silent SVC obstruction. Even though our patient had bilateral SVCs, there was no bridging vein, and hence she developed decompressing varices in response to left SVC occlusion in the absence of obstruction of the right SVC. Technical factors limited our ability to maintain patency of the recanalized left SVC. Unfortunately, alternate decompressing collaterals opened up, preventing recurrence of the esophageal varices. In adults, tumors are the most common cause of SVC obstruction and dominate the clinical picture (2); however, SVC obstruction in children is more likely secondary to vascular access complications or complex palliative procedures for congenital heart disease (3–5). Therefore, in children, bleeding from downhill varices may become the principal clinical presentation. This report brings to attention atypical features of bleeding esophageal varices secondary to SVC obstruction. Factors differentiating these from portal hypertensive varices are that they occur in the upper esophagus, drain in the downward (caudad) direction, are secondary to systemic venous obstruction rather than portal hypertension, and the treatment is aimed at relieving SVC obstruction. The increasing frequency with which SVC obstruction is being recognized warrants greater awareness of this silent but potentially serious clinical problem (5). In patients with known SVC obstruction, upper gastrointestinal endoscopy should be considered, even in the absence of obvious gastrointestinal hemorrhage. Similarly, the presence of upper esophageal varices should result in a search for SVC obstruction. Treatment options should target establishing patency of the SVC via interventional or surgical options.